In a diverse array of avian and mammalian species, experimental manipulations of clutch size have tested the hypothesis that natural selection should adjust numbers of neonates produced so as to maximize the number of viable offspring at the end of the period of parental care. Reptiles have not been studied in this respect, probably because they rarely display parental care. However, females of all python species brood their eggs until hatching, but they do not care for their neonates. This feature provides a straightforward way to experimentally increase or reduce clutch size to see whether the mean clutch size observed in nature does indeed maximize hatching success and/or optimize offspring phenotypes. Eggs were removed or added to newly laid clutches of Ball Pythons (Python regius) in tropical Africa (nine control clutches, eight with 50% more eggs added, six with 42% of eggs removed). All clutches were brooded by females throughout the 2-month incubation period. Experimental manipulation of clutch-size did not significantly affect the phenotypes (morphology, locomotor ability) of hatchlings, but eggs in 'enlarged' clutches hatched later, and embryos were more likely to die before hatching. This mortality was due to desiccation of the eggs, with females being unable to cover 'enlarged' clutches sufficiently to retard water loss. Our results support the notion of an optimal clutch size, driven by limitations on parental ability to care for the offspring. However, the proximate mechanisms that generate this optimum value differ from those previously described in other kinds of animals.